Fold recognition using predicted secondary structure sequences and hidden Markov models of protein folds

Fold-specific sequence scoring improves protein sequence matching

Background

Sequence matching is extremely important for applications throughout biology, particularly for discovering information such as functional and evolutionary relationships, and also for discriminating between unimportant and disease mutants. At present the functions of a large fraction of genes are unknown; improvements in sequence matching will improve gene annotations. Universal amino acid substitution matrices such as Blosum62 are used to measure sequence similarities and to identify distant homologues, regardless of the structure class. However, such single matrices do not take into account important structural information evident within the different topologies of proteins and treats substitutions within all protein folds identically. Others have suggested that the use of structural information can lead to significant improvements in sequence matching but this has not yet been very effective. Here we develop novel substitution matrices that include not only general sequence information but also have a topology specific component that is unique for each CATH topology. This novel feature of using a combination of sequence and structure information for each protein topology significantly improves the sequence matching scores for the sequence pairs tested. We have used a novel multi-structure alignment method for each homology level of CATH in order to extract topological information.

Results

We obtain statistically significant improved sequence matching scores for 73 % of the alpha helical test cases. On average, 61 % of the test cases showed improvements in homology detection when structure information was incorporated into the substitution matrices. On average z-scores for homology detection are improved by more than 54 % for all cases, and some individual cases have z-scores more than twice those obtained using generic matrices. Our topology specific similarity matrices also outperform other traditional similarity matrices and single matrix based structure methods. When default amino acid substitution matrix in the Psi-blast algorithm is replaced by our structure-based matrices, the structure matching is significantly improved over conventional Psi-blast. It also outperforms results obtained for the corresponding HMM profiles generated for each topology.

Conclusions

We show that by incorporating topology-specific structure information in addition to sequence information into specific amino acid substitution matrices, the sequence matching scores and homology detection are significantly improved. Our topology specific similarity matrices outperform other traditional similarity matrices, single matrix based structure methods, also show improvement over conventional Psi-blast and HMM profile based methods in sequence matching. The results support the discriminatory ability of the new amino acid similarity matrices to distinguish between distant homologs and structurally dissimilar pairs.

Electronic supplementary material

The online version of this article (doi:10.1186/s12859-016-1198-z) contains supplementary material, which is available to authorized users.

JPred4: a protein secondary structure prediction server

JPred4 (http://www.compbio.dundee.ac.uk/jpred4) is the latest version of the popular JPred protein secondary structure prediction server which provides predictions by the JNet algorithm, one of the most accurate methods for secondary structure prediction. In addition to protein secondary structure, JPred also makes predictions of solvent accessibility and coiled-coil regions. The JPred service runs up to 94 000 jobs per month and has carried out over 1.5 million predictions in total for users in 179 countries. The JPred4 web server has been re-implemented in the Bootstrap framework and JavaScript to improve its design, usability and accessibility from mobile devices. JPred4 features higher accuracy, with a blind three-state (α-helix, β-strand and coil) secondary structure prediction accuracy of 82.0% while solvent accessibility prediction accuracy has been raised to 90% for residues <5% accessible. Reporting of results is enhanced both on the website and through the optional email summaries and batch submission results. Predictions are now presented in SVG format with options to view full multiple sequence alignments with and without gaps and insertions. Finally, the help-pages have been updated and tool-tips added as well as step-by-step tutorials.

ASTRO-FOLD: a combinatorial and global optimization framework for Ab initio prediction of three-dimensional structures of proteins from the amino acid sequence

The field of computational biology has been revolutionized by recent advances in genomics. The completion of a number of genome projects, including that of the human genome, has paved the way toward a variety of challenges and opportunities in bioinformatics and biological systems engineering. One of the first challenges has been the determination of the structures of proteins encoded by the individual genes. This problem, which represents the progression from sequence to structure (genomics to structural genomics), has been widely known as the structure-prediction-in-protein-folding problem. We present the development and application of ASTRO-FOLD, a novel and complete approach for the ab initio prediction of protein structures given only the amino acid sequences of the proteins. The approach exhibits many novel components and the merits of its application are examined for a suite of protein systems, including a number of targets from several critical-assessment-of-structure-prediction experiments.

Using 3D Hidden Markov Models that explicitly represent spatial coordinates to model and compare protein structures

BACKGROUND

Hidden Markov Models (HMMs) have proven very useful in computational biology for such applications as sequence pattern matching, gene-finding, and structure prediction. Thus far, however, they have been confined to representing 1D sequence (or the aspects of structure that could be represented by character strings).

RESULTS

We develop an HMM formalism that explicitly uses 3D coordinates in its match states. The match states are modeled by 3D Gaussian distributions centered on the mean coordinate position of each alpha carbon in a large structural alignment. The transition probabilities depend on the spread of the neighboring match states and on the number of gaps found in the structural alignment. We also develop methods for aligning query structures against 3D HMMs and scoring the result probabilistically. For 1D HMMs these tasks are accomplished by the Viterbi and forward algorithms. However, these will not work in unmodified form for the 3D problem, due to non-local quality of structural alignment, so we develop extensions of these algorithms for the 3D case. Several applications of 3D HMMs for protein structure classification are reported. A good separation of scores for different fold families suggests that the described construct is quite useful for protein structure analysis.

CONCLUSION

We have created a rigorous 3D HMM representation for protein structures and implemented a complete set of routines for building 3D HMMs in C and Perl. The code is freely available from http://www.molmovdb.org/geometry/3dHMM, and at this site we also have a simple prototype server to demonstrate the features of the described approach.

Improved recognition of native-like protein structures using a family of designed sequences

The goal of the inverse protein folding problem is to identify amino acid sequences that stabilize a given target protein conformation. Methods that attempt to solve this problem have proven useful for protein sequence design. Here we show that the same methods can provide valuable information for protein fold recognition and for ab initio protein structure prediction. We present a measure of the compatibility of a test sequence with a target model structure, based on computational protein design. The model structure is used as input to design a family of low free energy sequences, and these sequences are compared with the test sequence by using a metric in sequence space based on nearest-neighbor connectivity. We find that this measure is able to recognize the native fold of a myoglobin sequence among different globin folds. It is also powerful enough to recognize near-native protein structures among non-native models.

Pcons: a neural-network-based consensus predictor that improves fold recognition

During recent years many protein fold recognition methods have been developed, based on different algorithms and using various kinds of information. To examine the performance of these methods several evaluation experiments have been conducted. These include blind tests in CASP/CAFASP, large scale benchmarks, and long-term, continuous assessment with newly solved protein structures. These studies confirm the expectation that for different targets different methods produce the best predictions, and the final prediction accuracy could be improved if the available methods were combined in a perfect manner. In this article a neural-network-based consensus predictor, Pcons, is presented that attempts this task. Pcons attempts to select the best model out of those produced by six prediction servers, each using different methods. Pcons translates the confidence scores reported by each server into uniformly scaled values corresponding to the expected accuracy of each model. The translated scores as well as the similarity between models produced by different servers is used in the final selection. According to the analysis based on two unrelated sets of newly solved proteins, Pcons outperforms any single server by generating approximately 8%-10% more correct predictions. Furthermore, the specificity of Pcons is significantly higher than for any individual server. From analyzing different input data to Pcons it can be shown that the improvement is mainly attributable to measurement of the similarity between the different models. Pcons is freely accessible for the academic community through the protein structure-prediction metaserver at http://bioinfo.pl/meta/.

A study of quality measures for protein threading models

BACKGROUND

Prediction of protein structures is one of the fundamental challenges in biology today. To fully understand how well different prediction methods perform, it is necessary to use measures that evaluate their performance. Every two years, starting in 1994, the CASP (Critical Assessment of protein Structure Prediction) process has been organized to evaluate the ability of different predictors to blindly predict the structure of proteins. To capture different features of the models, several measures have been developed during the CASP processes. However, these measures have not been examined in detail before. In an attempt to develop fully automatic measures that can be used in CASP, as well as in other type of benchmarking experiments, we have compared twenty-one measures. These measures include the measures used in CASP3 and CASP2 as well as have measures introduced later. We have studied their ability to distinguish between the better and worse models submitted to CASP3 and the correlation between them.

RESULTS

Using a small set of 1340 models for 23 different targets we show that most methods correlate with each other. Most pairs of measures show a correlation coefficient of about 0.5. The correlation is slightly higher for measures of similar types. We found that a significant problem when developing automatic measures is how to deal with proteins of different length. Also the comparisons between different measures is complicated as many measures are dependent on the size of the target. We show that the manual assessment can be reproduced to about 70% using automatic measures. Alignment independent measures, detects slightly more of the models with the correct fold, while alignment dependent measures agree better when selecting the best models for each target. Finally we show that using automatic measures would, to a large extent, reproduce the assessors ranking of the predictors at CASP3.

CONCLUSIONS

We show that given a sufficient number of targets the manual and automatic measures would have given almost identical results at CASP3. If the intent is to reproduce the type of scoring done by the manual assessor in in CASP3, the best approach might be to use a combination of alignment independent and alignment dependent measures, as used in several recent studies.

Use of residue pairs in protein sequence-sequence and sequence-structure alignments

Two new sets of scoring matrices are introduced: H2 for the protein sequence comparison and T2 for the protein sequence-structure correlation. Each element of H2 or T2 measures the frequency with which a pair of amino acid types in one protein, k-residues apart in the sequence, is aligned with another pair of residues, of given amino acid types (for H2) or in given structural states (for T2), in other structurally homologous proteins. There are four types, corresponding to the k-values of 1 to 4, for both H2 and T2. These matrices were set up using a large number of structurally homologous protein pairs, with little sequence homology between the pair, that were recently generated using the structure comparison program SHEBA. The two scoring matrices were incorporated into the main body of the sequence alignment program SSEARCH in the FASTA package and tested in a fold recognition setting in which a set of 107 test sequences were aligned to each of a panel of 3,539 domains that represent all known protein structures. Six procedures were tested; the straight Smith-Waterman (SW) and FASTA procedures, which used the Blosum62 single residue type substitution matrix; BLAST and PSI-BLAST procedures, which also used the Blosum62 matrix; PASH, which used Blosum62 and H2 matrices; and PASSC, which used Blosum62, H2, and T2 matrices. All procedures gave similar results when the probe and target sequences had greater than 30% sequence identity. However, when the sequence identity was below 30%, a similar structure could be found for more sequences using PASSC than using any other procedure. PASH and PSI-BLAST gave the next best results.